Published Paper Details:

There are multiple methods available for monitoring traffic conditions on roadways. With the advancement of artificial intelligence (AI) in image processing technology, there is an increasing interest in creating traffic monitoring systems that utilize camera vision data. This research paper presents a technique for extracting traffic information from a camera positioned at an intersection, aimed at enhancing road monitoring systems. The approach employs a deep learning model (YOLOv4) for processing images to detect vehicles and classify their types. Vehicle trajectories are estimated lane by lane by correlating the detected vehicle positions with a high definition map (HD map). From these estimated trajectories, the traffic volumes for each lane's direction of travel are calculated. The effectiveness of the proposed method was evaluated using samples across distinct criteria: vehicle detection rate, traffic volume estimation. The findings indicate a 99% success rate in vehicle detection, introduced in this research demonstrates the potential for gathering detailed traffic data installed at an intersection. The integration of AI technologies represents the primary contribution of this study, highlighting significant potential for enhancing existing traffic monitoring systems